I decided to have a rethink on the voltage and current sensing. First the current. I looked at the popular lem hass series of transducers but didnt like the limited output of only 625mv for full scale. Noise would be an issue and the code would need to be altered beyond my ability. Some searching revealed the Honeywell CSNF651 senseor:
http://radionics.rs-online.com/web/s...duct&R=6673465
This simple little guy outputs 50mA at 100A primary. So for a 50amp sense setup it just needs a 200R precision resistor to turn 25ma (50A) into a 5v signal and is of course variable and fully isolated.
Next the voltage. I had looked at linear optos but the only ones i could source were messy setups. Came upon the LEM LV25-p voltage transducer:
http://radionics.rs-online.com/web/s...duct&R=0286361
Similar to the current sensor it converts primary current to secondary current with full isolation. Nominal primary current of 10ma results in a secondary current of 25ma. So a primary resistor of 10k would result for a nominal voltage of 100v and a precision 200R secondary resistor converts this inot a 5v o/p. the only real disadvantage here is that the primary resistor must be chosen to be reasonably close to the nominal pack voltage being charged. 96V in my case so I use 100v for easy maths. This does bring a little added complexity but the resistor could be mounted in a sil socket for easy changes.
These two mods , I believe , solve the isolation issue. The control board gnd can then be tied to mains earth for complete safety. The only unisolated parts being inside an earthed metal enclosure makes it no more hazardous than a pc power supply. Granted the battery terminals are still "hot" but good practice would be to have the terminals fully insulated in any case. The charger is probably close to $300 with these changes but still exteremy cheap and versatile.
Soon as Radionics reopen next week (year?) i'll get the bits in and report on progress. Any thoughts?